Event determination method and device, computer equipment and storage medium

Abstract

The invention discloses an event determination method and device, computer equipment and a storage medium, and belongs to the technical field of computers. The method comprises the following steps: clustering a plurality of obtained texts to obtain a plurality of candidate text clusters; in the multiple candidate text clusters, the candidate text clusters with the number of texts meeting a first number condition are determined as target text clusters, and events associated with the multiple texts in the target text clusters belong to target type events; determining weights of the plurality of texts in the target text cluster based on the interaction information of the plurality of texts in the target text cluster; based on the multiple texts in the target text cluster and the weights of the multiple texts in the target text cluster, an event description text is generated, the semantics of the event description text is the same as the semantics of the multiple texts in the target text cluster, and the event description text is used for describing target type events associated with the multiple texts in the target text cluster. According to the method, computing resources can be saved, and the interpretability of the event determination method can be improved.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of computer technology, and particularly to an event determination method, apparatus, computer device, and storage medium. Background Art

[0002] With the rapid development of Internet technology, the Internet plays an increasingly important role in people's daily life, work, and study. Users can provide feedback on various products on the Internet, such as opinions, evaluations, or views on software functions or game play. Therefore, valuable references for subsequent product optimization and operation can be obtained by using the evaluation texts associated with products published on the Internet.

[0003] In related technologies, a pre-trained deep learning model is used for text classification, and then hot events are determined based on the classification results. A hot event refers to an event that is widely concerned and discussed, such as a bug (error) in a product. However, the deep learning-based processing method not only requires a large amount of computing resources but also has poor interpretability. Summary of the Invention

[0004] The embodiments of the present application provide an event determination method, apparatus, computer device, and storage medium, which can save computing resources and improve the interpretability of the event determination method. The technical solutions are as follows:

[0005] On the one hand, an event determination method is provided, and the method includes:

[0006] Clustering a plurality of obtained texts to obtain a plurality of candidate text clusters, each candidate text cluster including at least one text, where the text is a text associated with an event;

[0007] Among the plurality of candidate text clusters, a candidate text cluster whose text quantity meets a first quantity condition is determined as a target text cluster, and the events associated with the plurality of texts in the target text cluster belong to a target type event, where the target type event is an event whose associated text quantity meets the first quantity condition;

[0008] Based on the interaction information of the plurality of texts in the target text cluster, determine the weights of the plurality of texts in the target text cluster, where the weight of the text represents the importance degree of the text;

[0009] Based on the plurality of texts in the target text cluster and the weights of the plurality of texts in the target text cluster, generate an event description text, where the semantics of the event description text is the same as the semantics of the plurality of texts in the target text cluster, and the event description text is used to describe the target type event associated with the plurality of texts in the target text cluster.

[0010] On the other hand, an event determination device is provided, and the device includes:

[0011] A clustering module, configured to cluster a plurality of obtained texts to obtain a plurality of candidate text clusters, each candidate text cluster including at least one text, and the text is a text associated with an event;

[0012] A text cluster determination module, configured to determine, from the plurality of candidate text clusters, a candidate text cluster whose number of included texts meets a first quantity condition as a target text cluster, and events associated with multiple texts in the target text cluster belong to a target type event, where the target type event refers to an event whose associated text quantity meets the first quantity condition;

[0013] A weight determination module, configured to determine weights of multiple texts in the target text cluster based on interaction information of the multiple texts in the target text cluster, and the weight of the text represents the importance degree of the text;

[0014] A text generation module, configured to generate an event description text based on the multiple texts in the target text cluster and the weights of the multiple texts in the target text cluster, where the semantics of the event description text is the same as the semantics of the multiple texts in the target text cluster, and the event description text is used to describe the target type event associated with the multiple texts in the target text cluster.

[0015] Optionally, the clustering module is configured to:

[0016] Extract features from the plurality of texts to obtain first semantic features of each text;

[0017] When the dimension of the first semantic features of the plurality of texts is not greater than a preset dimension, cluster the plurality of texts based on the first semantic features of the plurality of texts to obtain the plurality of candidate text clusters;

[0018] When the dimension of the first semantic features of the plurality of texts is greater than the preset dimension, reduce the dimension of the first semantic features of the plurality of texts to obtain second semantic features of the plurality of texts, and cluster the plurality of texts based on the second semantic features of the plurality of texts to obtain the plurality of candidate text clusters.

[0019] Optionally, the clustering module is configured to:

[0020] Extract keywords from the plurality of texts respectively to obtain at least one keyword in each text;

[0021] Determine at least one first target keyword from the keywords in the plurality of texts, and the number of occurrences of the first target keyword in the plurality of texts meets a second quantity condition;

[0022] For any first target keyword, all texts including the first target keyword in the multiple texts are determined as a candidate text cluster.

[0023] Optionally, the clustering module is configured to:

[0024] Based on a plurality of preset topic tags, respectively identify the multiple texts to obtain the topic tag to which each text belongs;

[0025] Determine at least one target topic tag among the multiple topic tags, and the number of texts under the target topic tag satisfies a third quantity condition;

[0026] For any target topic tag, all texts belonging to the target topic tag in the multiple texts are determined as a candidate text cluster.

[0027] Optionally, the device further includes a text cluster merging module, configured to:

[0028] For each obtained candidate text cluster, among the keywords in the texts included in the candidate text cluster, determine a second target keyword of the candidate text cluster, and the number of occurrences of the second target keyword in the texts included in the candidate text cluster satisfies a fourth quantity condition;

[0029] When the second target keywords of two candidate text clusters are the same, merge the two candidate text clusters into one candidate text cluster.

[0030] Optionally, the device further includes a text cluster merging module, configured to:

[0031] Among the obtained multiple candidate text clusters, determine the text duplication degree between every two candidate text clusters, where the text duplication degree represents the proportion of the same texts in the two candidate text clusters;

[0032] When the text duplication degree between two candidate text clusters is not less than a first preset threshold, merge the two candidate text clusters into one candidate text cluster.

[0033] Optionally, the weight determination module is configured to:

[0034] For any text, obtain the interaction parameter in the interaction information of the text, where the interaction parameter includes at least one of the number of forwards, the number of likes, the number of comments, the number of views, the number of collections, or the number of shares of the text;

[0035] Perform a weighted sum of the interaction parameters of the text to obtain the weight of the text.

[0036] Optionally, the text generation module is configured to:

[0037] Determine multiple target texts from multiple texts in the target text cluster, where the weights of the target texts meet a preset weight condition;

[0038] Generate the event description text based on the multiple target texts, where the semantics of the event description text are the same as the semantics of the multiple target texts.

[0039] Optionally, the text generation module is configured to:

[0040] Obtain a prompt text, where the prompt text is used to describe the goals that the expected generated event description text needs to meet;

[0041] Input the multiple target texts and the prompt text into a language processing model to obtain the event description text output by the language processing model.

[0042] Optionally, the multiple texts are obtained within the same period, and the target type events associated with the multiple texts in the target text cluster are target type events that occur within the period; the apparatus further includes an event determination module, configured to:

[0043] Determine the similarity between a first target text cluster and a second target text cluster obtained in different periods;

[0044] In the case where the similarity between the first target text cluster and the second target text cluster is not less than a second preset threshold, determine that the target type events associated with the multiple texts in the first target text cluster and the target type events associated with the multiple texts in the second target text cluster belong to the same target type event.

[0045] Optionally, the event determination module is configured to:

[0046] Aggregate keywords in multiple texts in the first target text cluster to obtain at least one first aggregated keyword;

[0047] Aggregate keywords in multiple texts in the second target text cluster to obtain at least one second aggregated keyword;

[0048] Determine the similarity between the at least one first aggregated keyword and the at least one second aggregated keyword as the similarity between the first target text cluster and the second target text cluster.

[0049] Optionally, the apparatus further includes a display module, configured to:

[0050] In the event display interface, event information of multiple determined target type events is displayed;

[0051] Wherein, the event information includes at least one of the event description text, the text in the target text cluster, the number of texts in the target text cluster, the keywords in the texts in the target text cluster, the occurrence times of the keywords, the new flag, or the occurrence times of the target type event within a preset duration.

[0052] On the other hand, a computer device is provided. The computer device includes a processor and a memory. At least one computer program is stored in the memory. The at least one computer program is loaded and executed by the processor to implement the operations performed by the event determination method as described in the above aspect.

[0053] On the other hand, a computer-readable storage medium is provided. At least one computer program is stored in the computer-readable storage medium. The at least one computer program is loaded and executed by a processor to implement the operations performed by the event determination method as described in the above aspect.

[0054] On the other hand, a computer program product is provided, including a computer program. The computer program is loaded and executed by a processor to implement the operations performed by the event determination method as described in the above aspect.

[0055] In the solution provided by the embodiments of the present application, when determining a target type event based on multiple texts associated with the event, first, the multiple texts are clustered to obtain multiple candidate text clusters, and then the target text clusters that meet the first quantity condition in terms of the number of texts are screened out from the multiple candidate text clusters. Then, the event associated with the multiple texts in the target text cluster belongs to the target type event. Subsequently, the event description text of the event is generated by using the multiple texts in the target text cluster and the weights of the multiple texts. Since the importance of each text is considered, the generated event description text can be guaranteed to focus on important aspects. Therefore, although the processing method of the solution of the present application is simple and saves computing resources, the accuracy of the determined target type event can still be guaranteed. And by generating the event description text corresponding to the target text cluster, the user can clearly understand the target type event and the multiple texts reflecting the target type event, improving the interpretability of the event determination method. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0057] Figure 1 is a schematic diagram of an implementation environment provided by an embodiment of the present application;

[0058] Figure 2 is a flowchart of an event determination method provided by an embodiment of the present application;

[0059] Figure 3 is a flowchart of another event determination method provided by an embodiment of the present application;

[0060] Figure 4 is a schematic diagram of a text clustering provided by an embodiment of the present application;

[0061] Figure 5 is a flowchart of a target type event processing method provided by an embodiment of the present application;

[0062] Figure 6 is a schematic diagram of a target type event judgment method provided by an embodiment of the present application;

[0063] Figure 7 is a schematic diagram of an event display interface provided by an embodiment of the present application;

[0064] Figure 8 is a schematic diagram of another event display interface provided by an embodiment of the present application;

[0065] Figure 9 is a system architecture diagram of an event determination method provided by an embodiment of the present application;

[0066] Figure 10 is a structural schematic diagram of an event determination device provided by an embodiment of the present application;

[0067] Figure 11 is a structural schematic diagram of another event determination device provided by an embodiment of the present application;

[0068] Figure 12 is a structural schematic diagram of a terminal provided by an embodiment of the present application;

[0069] Figure 13 is a structural schematic diagram of a server provided by an embodiment of the present application. Detailed implementation manners

[0070] To make the objectives, technical solutions and advantages of the embodiments of the present application clearer, the following will further describe the implementation manners of the present application in detail with reference to the accompanying drawings.

[0071] It can be understood that the terms "first", "second", etc. used in this application may be used herein to describe various concepts, but unless otherwise specified, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of this application, the first target text cluster may be referred to as the second target text cluster, and similarly, the second target text cluster may be referred to as the first target text cluster.

[0072] Among them, at least one means one or more than one. For example, at least one text can be one text, two texts, three texts, etc., any integer greater than or equal to one. A plurality means two or more than two. For example, a plurality of texts can be two texts, three texts, etc., any integer greater than or equal to two. Each means every one of at least one. For example, each text means every one of a plurality of texts. If there are 3 texts in a plurality of texts, then each text means every one of the 3 texts.

[0073] It should be noted that the information (including but not limited to user device information, user personal information, etc.), data (including but not limited to data for analysis, stored data, displayed data, etc.), and signals (including but not limited to signals transmitted between user terminals and other devices, etc.) involved in this application are all fully authorized by users or relevant parties, and the collection, use, and processing of relevant data need to comply with relevant laws, regulations, and standards of relevant countries and regions. For example, the multiple texts involved in this application are all obtained under the premise that the user is informed in advance and fully authorized.

[0074] Artificial Intelligence (AI) is a theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology in computer science. It attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a way similar to human intelligence. Artificial intelligence also studies the design principles and implementation methods of various intelligent machines, enabling the machines to have the functions of perception, reasoning, and decision-making.

[0075] Artificial intelligence technology is an interdisciplinary subject that covers a wide range of fields, including both hardware-level and software-level technologies. The basic technologies of artificial intelligence generally include sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, pre-trained model technology, operating / interactive systems, mechatronics, etc. Among them, the pre-trained model, also known as the large model or the foundation model, can be widely applied to downstream tasks in various directions of artificial intelligence after fine-tuning. Artificial intelligence software technology mainly includes several major directions such as computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0076] Natural Language Processing (NLP) is an important direction in the fields of computer science and artificial intelligence. It studies various theories and methods that can achieve effective communication between humans and computers using natural language. Natural language processing involves natural language, that is, the language people use in daily life, and is closely related to linguistic research; at the same time, it involves important technologies in the fields of computer science and mathematics such as model training in the field of artificial intelligence. The pre-trained model (Pre-trained Models, PTM) is developed from the large language model in the NLP field. The pre-trained model, also known as the foundation model and the large model, refers to a deep neural network (Deep Neural Network, DNN) with a large number of parameters. It is trained on a large amount of unlabeled data, and the function approximation ability of the large-parameter DNN is used to enable the PTM to extract common features from the data. Through technologies such as fine-tuning, parameter-efficient fine-tuning (PEFT), and Prompt-Tuning (prompt tuning), it is applicable to downstream tasks. Therefore, the pre-trained model can achieve ideal results in few-shot or zero-shot scenarios. PTMs are classified into language models, visual models, speech models, multi-modal models, etc. according to the data modalities they process. For example, language models include ELMO (Embeddings from Language Model, a language model), BERT (Bidirectional Encoder Representations from Transformers, a bidirectional pre-trained language model), GPT (Generative Pre-trained Transformer, a pre-trained generative model), etc. Among them, the multi-modal model refers to a model that establishes feature representations of two or more data modalities. The pre-trained model is an important tool for outputting Artificial Intelligence Generated Content (AIGC) and can also be used as a general interface connecting multiple specific task models. After fine-tuning, the large language model can be widely applied to downstream tasks. Natural language processing technologies usually include text processing, semantic understanding, machine translation, robot question answering, knowledge graphs and other technologies.

[0077] Machine Learning (ML) is an interdisciplinary field that involves multiple disciplines such as probability theory, statistics, approximation theory, convex analysis, and algorithm complexity theory. It specifically studies how computers can simulate or implement human learning behaviors to acquire new knowledge or skills and reorganize the existing knowledge structure to continuously improve their own performance. Machine learning is the core of artificial intelligence and the fundamental way to make computers intelligent, and its applications cover all fields of artificial intelligence. Machine learning and deep learning usually include technologies such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and learning from demonstration. Pre-trained models are the latest development results of deep learning, integrating the above technologies.

[0078] The event determination method provided in the embodiments of the present application will be described below based on artificial intelligence technology and natural language processing technology.

[0079] The event determination method provided in the embodiments of the present application can be used in a computer device. Optionally, the computer device is a terminal or a server. Optionally, the server is an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms. Optionally, the terminal is a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, etc., but is not limited thereto.

[0080] In a possible implementation manner, the computer program involved in the embodiments of the present application can be deployed to be executed on a computer device, or on multiple computer devices located at one place, or on multiple computer devices distributed at multiple places and interconnected through a communication network. The multiple computer devices distributed at multiple places and interconnected through a communication network can form a blockchain system.

[0081] Figure 1 is a schematic diagram of an implementation environment provided in the embodiments of the present application. See Figure 1 , the implementation environment includes: a terminal 101 and a server 102. The terminal 101 and the server 102 are connected through a wireless or wired network.

[0082] Server 102 obtains multiple texts associated with an event, clusters the multiple texts to obtain multiple candidate text clusters, and filters out a target text cluster whose number of included texts meets the first quantity condition. The events associated with the multiple texts in the target text cluster belong to the target type of event, and the target type of event refers to an event whose associated number of texts meets the first quantity condition. Since the number of texts associated with the target type of event is large, that is, there are many texts discussing the target type of event, and thus the target type of event has a high popularity. Therefore, the target type of event can be understood as a hot event, that is, an event that is widely concerned and discussed. For example, for an electronic game application, the texts associated with the electronic game application include the texts published in the forum of the electronic game application, and these texts are obtained on the basis of the user's prior full knowledge and authorization. If the number of texts discussing the rendering bug in the electronic game application is large, it indicates that the event of the rendering bug in the electronic game application has a high popularity. Therefore, the event of the rendering bug is a hot event.

[0083] After determining the target text cluster, Server 102 generates an event description text based on the multiple texts in the target text cluster, and the event description text is used to describe the target type of event. Subsequently, Server 102 sends the target text cluster and the event description text to Terminal 101, and Terminal 101 displays the event description text and the target text cluster, so that the user can understand the target type of event and thus quickly understand the currently highly popular events.

[0084] In a possible implementation manner, an application client provided by the server runs on Terminal 101, and Server 102 sends the event description text and the target text cluster to the application client, and Terminal 101 displays the event description text and the target text cluster in the application client. Alternatively, Terminal 101 can also display the event description text and the target text cluster in the form of a web page, and the embodiments of the present application do not make any limitations in this regard.

[0085] It should be noted that the above only takes Server 102 determining the target text cluster and the corresponding event description text as an example for illustration. In another embodiment, Server 102 can also send the multiple texts associated with the event to Terminal 101, and Terminal 101 determines the target text cluster and the event description text based on the multiple texts. The embodiments of the present application do not make any limitations on the execution entity for determining the target text cluster and the event description text.

[0086] The event determination method provided by the embodiments of the present application can be applied to any scenario for determining the target type of event.

[0087] For example, the embodiments of the present application are applied to a scenario of determining a target type of event in the field of electronic games. Among them, the target type of event can be understood as a hot event in the field of electronic games. For a certain electronic game application, a computer device obtains text related to the electronic game application. For example, the text related to the electronic game application includes feedback text submitted through a specified feedback channel of the electronic game application. On the basis that the user is fully informed and authorized in advance, it may also include: text published in the discussion area of the electronic game application, text published in the comment area of the electronic game application, text published in the forum of the electronic game application, etc. The computer device uses the event determination method provided by the embodiments of the present application, and based on the obtained text, determines a target text cluster and a corresponding event description text, and the event description text can reflect a hot event related to the electronic game application. For example, the hot events include bugs appearing in the electronic game application, game competitions held in the electronic game application, new game characters launched in the electronic game application, etc.

[0088] In addition, the method provided by the embodiments of the present application can also be applied to other scenarios, and the embodiments of the present application do not limit the application scenarios.

[0089] Figure 2 is a flowchart of an event determination method provided by the embodiments of the present application, which is executed by a computer device. Refer to Figure 2 and the method includes:

[0090] 201. The computer device clusters a plurality of obtained texts to obtain a plurality of candidate text clusters, and each candidate text cluster includes at least one text, and the text is a text associated with an event.

[0091] The computer device obtains a plurality of texts associated with an event, and the text associated with the event refers to a text that can reflect the event. Optionally, the text is evaluation text for feedback on various products published on the Internet, such as opinions, evaluations, or views on software functions or game play.

[0092] Among them, the text and the event in the embodiments of the present application can belong to any field. For example, in the game field, the text associated with the event refers to evaluation text published by players for an electronic game, which can reflect the opinions, attitudes, and views of players on the electronic game.

[0093] After the computer device obtains a plurality of texts, it clusters the plurality of texts to obtain a plurality of candidate text clusters, and each candidate text cluster includes at least one text. It can be considered that one or more texts in a candidate text cluster are texts associated with the same event, that is, the texts in a candidate text cluster reflect the same event.

[0094] Among multiple candidate text clusters, the computer device determines as the target text cluster a candidate text cluster whose contained text quantity meets the first quantity condition, where the events associated with the multiple texts in the target text cluster belong to the target type of event, and the target type of event refers to an event whose associated text quantity meets the first quantity condition.

[0095] After determining multiple candidate text clusters, the computer device determines the quantity of the texts contained in each candidate text cluster, and determines as the target text cluster a candidate text cluster whose contained text quantity meets the first quantity condition. Among them, the quantity of texts in the target text cluster that meets the first quantity condition is greater than the quantity of texts in the candidate text clusters that do not meet the first quantity condition.

[0096] Optionally, the first quantity condition means that the quantity of texts contained in the text cluster reaches a first preset quantity, and the first preset quantity is a quantity set in advance and can be flexibly set according to the actual situation. For example, if the first preset quantity is 1000, then the first quantity condition means that the quantity of texts contained in the text cluster reaches 1000. Or, the first quantity condition means that in the order of the quantity of texts in the text cluster from largest to smallest, it is within the first second preset quantity. For example, if the second preset quantity is 20, then the first quantity condition means that in the order of the quantity of texts in the text cluster from largest to smallest, it is within the first 20. The embodiments of the present application do not limit the first quantity condition.

[0097] Among them, the target text cluster refers to a candidate text cluster whose contained text quantity meets the first quantity condition. Here, the word "target" can be understood as the contained text quantity meets the first quantity condition, that is, when the candidate text cluster reaches the goal of "the contained text quantity meets the first quantity condition", then this candidate text cluster can be determined as the target text cluster.

[0098] Among them, the target type of event refers to an event whose associated text quantity meets the first quantity condition. Here, the word "target" can be understood as the associated text quantity meets the first quantity condition, that is, when the event reaches the goal of "the associated text quantity meets the first quantity condition", then this event can be determined as the target type of event.

[0099] Since the target text cluster refers to a candidate text cluster whose contained text quantity meets the first quantity condition, and the target text cluster is obtained by clustering, it can be considered that the events associated with the texts in the target text cluster are the same event. And since the text quantity associated with this event meets the first quantity condition, this event belongs to the target type of event. Therefore, the events associated with the multiple texts in the target text cluster belong to the target type of event.

[0100] In the embodiments of the present application, the event reflected by the target text cluster belongs to the target type event. The number of texts associated with the target type event satisfies the first quantity condition. The number of texts associated with the target type event is relatively large, that is, there are more texts discussing the target type event. Furthermore, the popularity of the target type event is relatively high. Therefore, the target type event can be understood as a hot event, that is, an event that is widely concerned and discussed, and has the characteristic of high popularity. For example, in the game field, the hot event refers to that it is easy to have bugs in a certain game mode, or there are deficiencies in the gameplay of the newly launched game character, etc.

[0101] It should be noted that the number of target text clusters determined in the embodiments of the present application can be one or more.

[0102] 203. The computer device determines the weights of multiple texts in the target text cluster based on the interaction information of the multiple texts in the target text cluster. The weight of a text represents the importance degree of the text.

[0103] After the computer device determines the target text cluster, for any target text cluster, the computer device obtains the interaction information of multiple texts in the target text cluster, and determines the weights of the multiple texts respectively based on the interaction information of the multiple texts. The weight of the text represents the importance degree of the text. Among them, the more active the interaction situation reflected by the interaction information of the text is, the higher the importance degree of the text is. The less active the interaction situation reflected by the interaction information of the text is, the lower the importance degree of the text is.

[0104] 204. The computer device generates an event description text based on multiple texts in the target text cluster and the weights of multiple texts in the target text cluster. The semantics of the event description text is the same as the semantics of multiple texts in the target text cluster. The event description text is used to describe the target type event associated with multiple texts in the target text cluster.

[0105] After the computer device determines the weights of multiple texts in the target text cluster, it generates an event description text based on multiple texts in the target text cluster and the weights of multiple texts in the target text cluster. The semantics of the event description text is the same as the semantics of multiple texts in the target text cluster. It can be understood that the event description text is a summary of the multiple texts. Subsequently, by viewing the event description text, one can quickly understand what target type event has occurred.

[0106] Among them, the greater the weight of the text is, the greater the influence degree of the text on the generated event description text is. The smaller the weight of the text is, the smaller the influence degree of the text on the generated event description text is.

[0107] The method provided by the embodiments of the present application, when determining a target type event based on multiple texts associated with an event, first clusters the multiple texts to obtain multiple candidate text clusters, then filters out a target text cluster whose text quantity meets the first quantity condition from the multiple candidate text clusters. The events associated with the multiple texts in the target text cluster belong to the target type event. Then, using the multiple texts in the target text cluster and the weights of the multiple texts, an event description text of the event is generated. Since the importance degree of each text is considered, it can be ensured that the generated event description text focuses on important aspects. Therefore, although the processing method of the solution of the present application is simple and saves computing resources, the accuracy of the determined target type event can still be ensured. And by generating the event description text corresponding to the target text cluster, the user can clearly understand the target type event and the multiple texts reflecting the target type event, improving the interpretability of the event determination method.

[0108] Figure 3 is a flowchart of another event determination method provided by the embodiments of the present application. The embodiments of the present application are executed by a computer device. Refer to Figure 3 and the method includes:

[0109] 301. The computer device clusters the multiple texts obtained to obtain multiple candidate text clusters. Each candidate text cluster includes at least one text, and the text is a text associated with an event.

[0110] In a possible implementation manner, when the computer device clusters the multiple texts obtained to obtain multiple candidate text clusters, it includes: obtaining multiple candidate texts, filtering out the texts that meet the valid conditions from the multiple candidate texts, and clustering the multiple texts that meet the valid conditions to obtain multiple candidate text clusters.

[0111] That is, the computer device first preprocesses the multiple candidate texts obtained, filters the multiple candidate texts using the valid conditions, and subsequently only uses the texts that meet the valid conditions and are filtered out for clustering.

[0112] Among them, the valid conditions include at least one of the following: (1) The length of the candidate text is not less than the preset length. If the length of the candidate text is less than the preset length, it is considered that the candidate text is not sufficient to reflect an event, so it is considered that the candidate text does not meet the valid conditions, and thus the candidate text is filtered out. (2) The candidate text includes at least one character. If the candidate text does not include characters, for example, the candidate text is all composed of special symbols or emoticons, then the candidate text has no substantial meaning and is not sufficient to reflect an event, so it is considered that the candidate text does not meet the valid conditions, and thus the candidate text is filtered out.

[0113] In a possible implementation, clustering multiple texts by a computer device includes three clustering methods, and the computer device can adopt at least one of the following three clustering methods.

[0114] The first clustering method: The computer device extracts features from multiple texts to obtain the first semantic feature of each text; when the dimension of the first semantic features of the multiple texts is not greater than a preset dimension, based on the first semantic features of the multiple texts, cluster the multiple texts to obtain multiple candidate text clusters; when the dimension of the first semantic features of the multiple texts is greater than the preset dimension, reduce the dimension of the first semantic features of the multiple texts to obtain the second semantic features of the multiple texts, and based on the second semantic features of the multiple texts, cluster the multiple texts to obtain multiple candidate text clusters.

[0115] Considering the diversity of texts and semantic features, if different feature extraction methods are used to extract features from texts, the dimensions of the obtained semantic features will be different. However, if the dimension of the semantic features is relatively high, the semantic features will be relatively sparse and not easy to cluster. Therefore, it is divided into two cases here. If the dimension of the first semantic feature obtained by feature extraction is not greater than the preset dimension, directly cluster based on the first semantic feature. If the dimension of the first semantic feature obtained by feature extraction is greater than the preset dimension, first reduce the dimension of the first semantic feature, and use the second semantic feature obtained after dimension reduction for clustering. Optionally, the dimension of the second semantic feature is not greater than the preset dimension.

[0116] Optionally, the computer device extracts features from multiple texts through a feature extraction model to obtain the first semantic feature. For example, the feature extraction model is an AI model suitable for text clustering tasks in the SentenceTransformer (text conversion) library. If the AI model is an English model, when the text is not in English, first translate the text into English, and then input it into the AI model for feature extraction.

[0117] Optionally, the computer device uses a non-linear dimensionality reduction technique (UMAP, Uniform Manifold Approximation and Projection) to reduce the dimension of the first semantic feature.

[0118] Optionally, the computer device uses the HDBSCAN (Hierarchical Density-Based Spatial Clustering of Applications with Noise) method for clustering, or uses the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) method for clustering.

[0119] Optionally, the computer device clusters multiple first semantic features or multiple second semantic features to obtain multiple candidate feature clusters. For any candidate feature cluster, the texts corresponding to the multiple semantic features in the candidate feature cluster are formed into a candidate text cluster, thereby obtaining multiple candidate text clusters.

[0120] Figure 4 It is a schematic diagram of a text clustering provided by an embodiment of the present application. As Figure 4 shown, the computer device represents multiple texts 401 using text embedding (that is, feature extraction) to obtain first semantic features 402. If the dimension of the multiple first semantic features 402 is not greater than the preset dimension, then the multiple first semantic features 402 are clustered to obtain multiple candidate feature clusters 404, and the multiple candidate feature clusters 404 are converted into multiple candidate text clusters 405. If the dimension of the multiple first semantic features 402 is greater than the preset dimension, then the multiple first semantic features 402 are dimensionally reduced to obtain multiple second semantic features 403, the multiple second semantic features 403 are clustered to obtain multiple candidate feature clusters 404, and the multiple candidate feature clusters 404 are converted into multiple candidate text clusters 405.

[0121] In the embodiment of the present application, semantic features are obtained by performing feature extraction on texts, and semantic clustering is performed using the semantic features. And if the dimension of the semantic features is greater than the preset dimension, then the semantic features are dimensionally reduced before performing semantic clustering to ensure that the semantic features used for semantic clustering are low-dimensional features, avoiding the problem of poor clustering effect caused by using high-dimensional features with sparse features, and improving the clustering accuracy of the multiple candidate text clusters obtained.

[0122] The second clustering method: The computer device extracts keywords from multiple texts respectively to obtain at least one keyword in each text; at least one first target keyword is determined among the keywords in the multiple texts, and the number of occurrences of the first target keyword in the multiple texts satisfies the second quantity condition; for any first target keyword, all the texts including the first target keyword in the multiple texts are determined as a candidate text cluster.

[0123] Among them, the keywords in the text refer to the words with relatively strong relevance to the event. For example, in the game field, the keywords refer to the words related to the game in the text. For example, if the text is "The characters in this game are really ugly", then the keyword in this text is "character". Optionally, the computer device identifies the entities in the text through the NER (Name Entity Recognition) model to obtain candidate keywords. The computer device identifies the part-of-speech of the words in the text through the POS-TAG (Part-of-Speech Tagging) model, and uses the words whose part-of-speech meets the conditions as candidate keywords according to the preset rules. Then, the candidate keywords are filtered using the preset blacklist to obtain the keywords in the text.

[0124] After the computer device obtains the keywords of each text, it aggregates the keywords in multiple texts to obtain multiple keywords, and determines the keywords that meet the second quantity condition as the first target keywords among the multiple keywords. Among them, the number of keywords that meet the second quantity condition is greater than the number of keywords that do not meet the second quantity condition. Optionally, the second quantity condition means that the number of keywords reaches the third preset quantity, or the second quantity condition means that they are arranged in descending order of the number of keywords, and are within the first four preset quantities in the front row. The embodiments of the present application do not limit the second quantity condition.

[0125] For each first target keyword, the computer device determines the texts including the first target keyword, and forms a candidate text cluster with all the texts including the first target keyword. Therefore, the number of candidate text clusters determined based on the keywords is equal to the number of first target keywords, and one first target keyword can determine one candidate text cluster.

[0126] The third clustering method: The computer device respectively identifies multiple texts based on a preset multiple topic tags to obtain the topic tags to which each text belongs; determines at least one target topic tag among the multiple topic tags, and the number of texts under the target topic tag meets the third quantity condition; for any target topic tag, determines all the texts belonging to the target topic tag among the multiple texts as a candidate text cluster.

[0127] Among them, the multiple topic tags are preset, and the topic tags can reflect the theme of the text. For example, in the game field, the multiple topic tags are set according to the knowledge of the game field. For example, if the text is "The characters in this game are really ugly", then the topic tag to which this text belongs is "character appearance".

[0128] Optionally, each topic tag corresponds to reference information, which includes reference words and reference text. The reference words and reference text belong to the topic tag, and the reference information can be understood as an example belonging to the topic tag. For any text, match the text with each piece of reference information to obtain the reference information with the highest similarity, and determine the topic tag corresponding to the reference information with the highest similarity as the topic tag to which the text belongs. Optionally, the step of matching the text with the reference information can be implemented by an AI model, such as a language processing model like a large language model, such as the GPT (Generative Pre-Trained Transformer) model.

[0129] After the computer device obtains the topic tag to which each text belongs, it determines the topic tag that meets the third quantity condition among the multiple topic tags. Among them, the number of texts corresponding to the topic tag that meets the third quantity condition is greater than the number of texts corresponding to the topic tag that does not meet the third quantity condition. Optionally, the third quantity condition means that the number of texts corresponding to the topic tag reaches a fifth preset quantity, or the third quantity condition means that in the order of the number of texts corresponding to the topic tags from largest to smallest, it is within the first six preset quantities. The embodiments of the present application do not limit the third quantity condition.

[0130] For each target topic tag, the computer device determines the texts belonging to the target topic tag, and forms a candidate text cluster with all the texts belonging to the target topic tag. Therefore, the number of candidate text clusters determined based on the topic tags is equal to the number of target topic tags, and one target topic tag can determine one candidate text cluster.

[0131] It should be noted that the above first clustering method belongs to clustering based on semantics, the second clustering method belongs to clustering based on keywords, and the third clustering method belongs to clustering based on topic tags. The above three clustering methods can be combined in pairs or all three. For example, using the above three methods for clustering at the same time, on the basis of clustering based on semantics, it is also combined with keywords and topic tags for clustering, which makes up for the deficiencies in the results of clustering based on semantics, improves the accuracy and comprehensiveness of the candidate text clusters obtained by clustering, and further improves the accuracy and comprehensiveness of the determined target type events.

[0132] In the embodiments of the present application, considering that the sources of the obtained texts may be different, there are significant differences in the styles of the respective texts, resulting in multiple actually similar texts being divided into multiple candidate text clusters. Moreover, if the above three different clustering methods are used for clustering simultaneously, it will also cause multiple actually similar texts to be divided into multiple candidate text clusters. Therefore, after obtaining multiple candidate text clusters, the candidate text clusters can be merged by performing at least one of the following steps 302 or 303, so as to avoid duplicate candidate text clusters, which is conducive to improving the accuracy of the determined target type events.

[0133] 302. For each obtained candidate text cluster, among the keywords in the texts included in the candidate text cluster, determine the second target keyword of the candidate text cluster, where the number of occurrences of the second target keyword in the texts included in the candidate text cluster satisfies the fourth quantity condition; in the case where the second target keywords of two candidate text clusters are the same, merge the two candidate text clusters into one candidate text cluster.

[0134] After the computer device determines multiple candidate text clusters, for each candidate text cluster, determine the keyword that satisfies the fourth quantity condition among the keywords of the texts in the candidate text cluster as the second target keyword of the candidate text cluster. Among them, the number of occurrences of the keyword that satisfies the fourth quantity condition is greater than the number of occurrences of the keyword that does not satisfy the fourth quantity condition. Optionally, the fourth quantity condition means that the number of occurrences of the keyword reaches a preset number, or the fourth quantity condition means that in the order of the number of occurrences of the keywords from largest to smallest, it is within the seventh preset quantity. The embodiments of the present application do not limit the fourth quantity condition.

[0135] If the second target keywords of two candidate text clusters are the same, it is considered that the events reflected by the two candidate text clusters are the same event. Therefore, the two candidate text clusters are merged into one candidate text cluster.

[0136] In a possible implementation manner, if the second target keywords of two candidate text clusters are the same, the two candidate text clusters are merged into one candidate text cluster, and text deduplication is performed on the merged candidate text cluster.

[0137] In a possible implementation manner, each candidate text cluster includes multiple second target keywords, and the same second target keywords of two candidate text clusters means that all the multiple second target keywords are the same.

[0138] 303. The computer device determines the text duplication degree between every two candidate text clusters among the obtained multiple candidate text clusters. The text duplication degree represents the proportion of the same text in the two candidate text clusters. When the text duplication degree between two candidate text clusters is not less than the first preset threshold, the two candidate text clusters are merged into one candidate text cluster.

[0139] After obtaining multiple candidate text clusters, the computer device determines the text duplication degree between every two candidate text clusters. If the text duplication degree between two candidate text clusters is not less than the first preset threshold, it is considered that the events reflected by these two candidate text clusters are the same event. Therefore, these two candidate text clusters are merged into one candidate text cluster.

[0140] In a possible implementation manner, each text corresponds to a unique text identifier. The computer device uses the text identifiers of the texts to determine whether the texts in two candidate text clusters are the same. For example, if there is a text with the text identifier A in one candidate text cluster and there is also a text with the text identifier A in another candidate text cluster, then the two candidate text clusters include the text indicated by the same text identifier A.

[0141] In a possible implementation manner, the text duplication degree between two candidate text clusters is equal to the ratio between the number of the same texts in the two candidate text clusters and the total number of texts in the two candidate text clusters. Optionally, the first preset threshold is equal to 70%.

[0142] It should be noted that the embodiments of this application are only described by taking the execution of the above steps 302 and 303 as examples. In another embodiment, the above steps 302 and 303 may not be executed, or only one of the above steps 302 and 303 may be executed.

[0143] 304. Among the multiple candidate text clusters, the computer device determines the candidate text clusters whose included text quantity meets the first quantity condition as target text clusters. The events associated with the multiple texts in the target text clusters belong to the target type events, and the target type events refer to the events whose associated text quantity meets the first quantity condition.

[0144] After determining the multiple candidate text clusters, the computer device determines the quantity of the texts included in each candidate text cluster, and determines the candidate text clusters whose included text quantity meets the first quantity condition as target text clusters. For example, arranged in descending order according to the included text quantity, the first 20 candidate text clusters are determined as target text clusters.

[0145] 305. Based on the interaction information of the multiple texts in the target text clusters, the computer device determines the weights of the multiple texts in the target text clusters. The weight of a text represents the importance degree of the text.

[0146] After the computer device determines the target text cluster, for any target text cluster, the computer device obtains the interaction information of multiple texts in the target text cluster, and based on the interaction information of the multiple texts, determines the weights of the multiple texts respectively. The weight of the text represents the importance of the text.

[0147] In a possible implementation, for any text, the computer device obtains the interaction parameters in the interaction information of the text, performs weighted summation on the interaction parameters of the text, and obtains the weight of the text.

[0148] Among them, the interaction parameters include at least one of the number of times the text is forwarded, the number of times the text is liked, the number of times the text is commented, the number of times the text is viewed, the number of times the text is collected, or the number of times the text is shared.

[0149] 306. The computer device determines multiple target texts among the multiple texts in the target text cluster, and the weights of the target texts meet the preset weight conditions.

[0150] After the computer device determines the weight of each text in the target text cluster, among the multiple texts in the target text cluster, the text whose weight meets the preset weight condition is determined as the target text. Among them, the weight of the text that meets the preset weight condition is greater than the weight of the text that does not meet the preset weight condition.

[0151] Optionally, the preset weight condition means that the weight reaches the preset weight threshold, or the preset weight condition means that in the order of weights from large to small, it is within the first eight preset quantities. The embodiments of the present application do not limit the preset weight conditions.

[0152] 307. The computer device generates an event description text based on the multiple target texts, and the semantics of the event description text are the same as the semantics of the multiple target texts.

[0153] In the embodiments of the present application, based on the weights of the texts, multiple target texts with higher importance are screened out. The computer device generates an event description text based on the multiple target texts, and the event description text is used to describe the target type event associated with the multiple texts in the target text cluster. Subsequently, by viewing the event description text, it is possible to quickly understand what event the target text cluster reflects, improving the readability of the target text cluster.

[0154] In a possible implementation, the computer device obtains a prompt text, and the prompt text is used to describe the target that the expected generated event description text needs to meet. The computer device inputs the multiple target texts and the prompt text into a language processing model, and obtains the event description text output by the language processing model. Optionally, the language processing model can be a GPT model, etc.

[0155] Among them, the prompt text can assist the language processing model to obtain an event description text that meets the user's expectations. The event description text can be understood as a summary of multiple target texts. Then, the prompt text can be: Use Chinese to summarize the following comment text, requiring it to be summarized into a coherent sentence in the form of stating the event, and the number of words should be controlled within 130, and try not to have emotional color, and allow rewriting to make the sentence more fluent.

[0156] In the embodiment of the present application, by executing the above-mentioned steps 306 and 307, it is realized that the event description text is generated based on the weights of multiple texts in the target text cluster and the multiple texts in the target text cluster. In another embodiment, other methods can also be adopted to realize the generation of event description text based on the weights of multiple texts in the target text cluster and the multiple texts in the target text cluster. For example, the computer device inputs the prompt text, multiple texts and the weights of multiple texts into the language processing model to obtain the event description text output by the language processing model. The degree of dependence of the generation process of the event description text on the input text depends on the weight of the text.

[0157] The method provided by the embodiment of the present application, when determining a target type event based on multiple texts associated with an event, first clusters the multiple texts to obtain multiple candidate text clusters, and then screens out a target text cluster whose number of texts meets a first quantity condition from the multiple candidate text clusters. Then, the event associated with the multiple texts in the target text cluster belongs to the target type event, and then uses the multiple texts in the target text cluster and the weights of the multiple texts to generate an event description text for the event. Since the importance of each text is taken into account, it can ensure that the generated event description text focuses on important angles. Therefore, although the processing method of the solution of the present application is simple and saves computing resources, it can still ensure the accuracy of the determined target type event, and by generating the event description text corresponding to the target text cluster, the user can clearly understand the target type event and the multiple texts reflecting the target type event, thereby improving the interpretability of the event determination method.

[0158] Moreover, the embodiment of the present application can automatically process a large amount of acquired text and can dynamically adapt to changes in the text without excessive manual intervention, thus saving a lot of manual work. At the same time, since there is no need to limit the event content of the target type event, even new target type events or events with increasing popularity can be discovered in a timely manner, thus improving the accuracy of determining the target type event.

[0159] In one possible implementation, the above Figure 3If multiple texts in the embodiments are obtained within the same period, then the target type events associated with the multiple texts in the target text cluster are target type events that occur within the period. Therefore, it is also possible to determine whether the target type events that occur in different periods belong to the same target type event, so as to conduct long-term observation on the same target type event. For the detailed process, please refer to the following Figure 5 embodiments. Figure 5 is a flowchart of a method for processing target type events provided by an embodiment of the present application. The embodiments of the present application are executed by a computer device. Please refer to Figure 5 , and the method includes:

[0160] 501. For the first target text cluster and the second target text cluster obtained by the computer device in different periods, determine the similarity between the first target text cluster and the second target text cluster.

[0161] In the embodiments of the present application, the computer device periodically executes the process in the above Figure 3 shown embodiments, so as to obtain target text clusters in multiple consecutive periods. Optionally, the period is one day, then the computer device executes the process in the above Figure 3 shown embodiments every other day. For example, at 24:00 every day, the computer device obtains the texts collected from 0:00 to 24:00, and determines the target text cluster and the corresponding event description text based on the collected texts. Optionally, the period can also be one week or one month, etc. The embodiments of the present application do not make any limitations in this regard.

[0162] In a possible implementation manner, the period to which the first target text cluster belongs is adjacent to the period to which the second target text cluster belongs. For example, the first target text cluster belongs to the first period, the second target text cluster belongs to the second period, and the first period is the previous period of the second period. Whenever the computer device determines the target text cluster and the corresponding event description text in the current period, it determines the similarity between each target text cluster in the current period and each target text cluster in the previous period, that is, for any two target text clusters being currently judged, the computer device executes the process of steps 501-step 502 in the embodiments of the present application for these two target text clusters.

[0163] In a possible implementation manner, the computer device uses the following steps 5021-step 5023 to determine the similarity between the first target text cluster and the second target text cluster.

[0164] 5021. The computer device aggregates the keywords in multiple texts in the first target text cluster to obtain at least one first aggregated keyword.

[0165] Optionally, the computer device filters the keywords in multiple texts in the first target text cluster based on the keyword blacklist to obtain multiple third target keywords, and aggregates them in the way of union-find based on the semantic similarity between the third target keywords to obtain at least one first aggregated keyword. Among them, if the semantic similarity between two third target keywords is greater than the third preset threshold, these two third target keywords are regarded as aggregatable, and the third target keyword with the highest word frequency among the two third target keywords is regarded as the first aggregated keyword obtained after aggregation.

[0166] Among them, the union-find is a data structure in computer science, mainly used to handle some problems about "sets", such as connectivity problems or set partitioning problems, etc. The union-find represents the elements in the set through an array structure, and each element points to another element, thus forming a network structure.

[0167] Optionally, the semantic similarity between the third target keywords is the dot product similarity between the word vectors of the third target keywords. For example, for the third target keyword w ori and the third target keyword w tgt , the computer device uses the following formula to determine the semantic similarity.

[0168]

[0169] Among them, sim(h ori ,h tgt ) represents the semantic similarity between the third target keyword w ori and the third target keyword w tgt , h ori represents the word vector of the third target keyword w ori , h tgt represents the word vector of the third target keyword w tgt . T represents transpose.

[0170] 5022. The computer device aggregates the keywords in multiple texts in the second target text cluster to obtain at least one second aggregated keyword.

[0171] Among them, the process of the computer device obtaining the second aggregated keyword is the same as the process of obtaining the first aggregated keyword in step 5021 above, and will not be elaborated here.

[0172] 5023. The computer device determines the similarity between at least one first aggregated keyword and at least one second aggregated keyword as the similarity between the first target text cluster and the second target text cluster.

[0173] The computer device determines the similarity between at least one first aggregated keyword and at least one second aggregated keyword, and determines this similarity as the similarity between the first target text cluster and the second target text cluster.

[0174] Optionally, the computer device determines the similarity using the following formula.

[0175]

[0176]

[0177] sim(E (ori) ,E (tgt) )=a1·sim ori (E (ori) ,E (tgt) )+a2·sim tgt (E (ori) ,E (tgt) );

[0178] Wherein, sim ori (E (ori) ,E (tgt) ) represents the first similarity, sim tgt (E (ori) ,E (tgt) ) represents the second similarity, E (ori) represents the first aggregated keyword set, E (tgt) represents the second aggregated keyword set, len(E (ori) ) represents the number of first aggregated keywords, len(E (tgt) represents the number of second aggregated keywords. represents the occurrence times of the i-th first aggregated keyword, represents the occurrence times of the j-th second aggregated keyword. represents the word vector of the i-th first aggregated keyword, represents the word vector of the j-th second aggregated keyword, represents the similarity between the word vector of the i-th first aggregated keyword and the word vector of the j-th second aggregated keyword. a1 and a2 are weights, and a1 and a2 can be equal to 0.5.

[0179] In the embodiments of the present application, when calculating the similarity using the above formula, a two-way weighting of the similarity between at least one first aggregated keyword and at least one second aggregated keyword is realized, which is beneficial to improving the accuracy of the determined similarity.

[0180] Figure 6 is a schematic diagram of a method for judging a target type event provided by an embodiment of the present application, as Figure 6As shown in the figure, the target text cluster of the target type event A in cycle 1 includes multiple original keywords. After filtering and aggregating the multiple original keywords, 2 aggregated keywords are obtained. The target text cluster of the target type event B in cycle 2 includes multiple original keywords. After filtering and aggregating the multiple original keywords, 4 aggregated keywords are obtained. Calculate the similarity between the 2 aggregated keywords of the target type event A and the 4 aggregated keywords of the target type event B. For example, the similarity is 0.6, and this similarity can be used as the similarity between the target text cluster of the target type event A and the target text cluster of the target type event B.

[0181] 502. When the similarity between the first target text cluster and the second target text cluster of the computer device is not less than the second preset threshold, it is determined that the target type events associated with the multiple texts in the first target text cluster and the target type events associated with the multiple texts in the second target text cluster belong to the same target type event.

[0182] If the similarity between the first target text cluster and the second target text cluster is not less than the second preset threshold, it is considered that the target type events reflected by the first target text cluster and the target type events reflected by the second target text cluster belong to the same target type event. Subsequently, the same event identifier can be set for the two target type events to indicate that the two target type events belong to the same target type event.

[0183] In the embodiments of the present application, considering that the discussion of a certain target type event often lasts for multiple days, in order to more intuitively present the volume change of the target type event, a recursive event re-identification method is provided to re-identify the target type events in different cycles, so as to obtain the same target type events in different cycles and assign the same event identifier. This method not only facilitates quickly understanding which target type events are new target type events, but also facilitates continuously observing the existing target type events to determine the volume evolution of the existing target type events. In addition, it is also convenient to determine the change in the view tendency for the same target type event.

[0184] 503. The computer device displays the event information of the determined multiple target type events in the event display interface.

[0185] Among them, each target text cluster corresponds to a target type event. After the computer device determines the target type event, it creates the event information of the target type event. Subsequently, in the event display interface, the event information of the determined multiple target type events can be displayed, which is convenient for users to quickly understand the target type events.

[0186] Among them, the event information includes at least one of an event description text, texts in a target text cluster, the number of texts in the target text cluster, keywords in the texts in the target text cluster, the occurrence times of the keywords, a new flag, or the occurrence times of a target type event within a preset time period. Optionally, the event information further includes a trend graph of the target type event, where the trend graph includes a first graph and a second graph. The first graph reflects the change in the extraction times of the target type event within the preset time period, and the second graph reflects the change in the popularity ranking of the target type event within the preset time period. For example, the first graph and the second graph are line graphs, curve graphs, or bar graphs, etc.

[0187] Figure 7 is a schematic diagram of an event display interface provided by an embodiment of the present application. As Figure 7 shown, the target type event can be referred to as a hot event. The time range, language range, and opinion tendency range to which the hot event to be displayed belongs can be set on the upper side of the event display interface, or new filtering conditions can be added. The event information of the hot event is displayed on the left side of the event display interface, including the event description text, keywords (such as word a, word b, etc.), the occurrence times of the keywords (such as word a appears 31 times, word b appears 20 times, etc.), the number of texts in the target text cluster (such as the number of texts in the target text cluster of the first hot event is 123), and a new identifier (such as "new" is displayed on the right side of the title of the second hot event). The language, source country, and trend graph corresponding to the currently selected hot event are displayed on the upper right side of the event display interface. The language refers to the language to which the texts in the target text cluster corresponding to the hot event belong, and the source country refers to the country from which the texts in the target text cluster corresponding to the hot event originate. The trend graph includes a line graph of the mention times of the hot event and a line graph of the popularity ranking. The texts in the target text cluster corresponding to the currently selected hot event are displayed on the lower right side of the event display interface.

[0188] Figure 8 is another schematic diagram of an event display interface provided by an embodiment of the present application. As Figure 8As shown, the target type event can be called a hot event. Taking the game field as an example, the hot events displayed in the event display interface belong to the hot events in the game field. On the left side of the event display interface, there are hot events "Online bug" and "Chat function not perfect". The mention count of the hot event "Online bug" is 13, and the mention count of the hot event "Chat function not perfect" is 10. The hot event "Online bug" is a newly emerged hot event, and its keywords include "bug", "online", "clipping", and "lag", etc. Below the hot event "Online bug", there is event description text "Players generally report that there are bugs in the online mode, including clipping, lag, etc., resulting in a poor game experience". On the lower right side of the event display interface, there are various texts corresponding to the currently selected hot event "Online bug". Figure 8 The event display interface in Figure 7 is the same as the above-mentioned

[0189] The method provided by the embodiment of the present application can display the event information of the target type event in the event display interface, which can not only intuitively display the keywords and occurrence times corresponding to each target type event, making the determined target type event have better interpretability, but also can display the event description text of the target type event, facilitating users to quickly understand the target type event, so as to analyze it in time and make corresponding decisions.

[0190] Figure 9 is the system architecture diagram of an event determination method provided by the embodiment of the present application. As Figure 9 shown, the method includes steps such as text preprocessing, text clustering, text cluster post-processing, target type event extraction, and event merging. The computer device obtains multiple candidate texts 901, performs text preprocessing on the multiple candidate texts 901 to obtain multiple preprocessed texts 902. The computer device performs text clustering on the multiple texts 902 to obtain multiple candidate text clusters 903. The computer device performs text cluster post-processing on the multiple candidate text clusters 903 to screen out multiple target text clusters 904. The computer device performs target type event extraction based on the multiple target text clusters 904 to obtain multiple event description texts, so as to describe the multiple target type events 905. The computer device performs event merging on the target type events 905 in different periods to determine the same target type events in different periods. For example, as Figure 9 shown, period 1 includes target type event A and target type event B, and period 2 includes target type events C and D. After calculation, it can be determined that target type event B and target type event D belong to the same event, and then these two target type events are given the same event identifier.

[0191] Figure 10The figure is a schematic structural diagram of an event determination device provided by an embodiment of the present application. Refer to Figure 10 , the device includes:

[0192] A clustering module 1001, configured to cluster a plurality of obtained texts to obtain a plurality of candidate text clusters, each candidate text cluster includes at least one text, and the text is a text associated with an event;

[0193] A text cluster determination module 1002, configured to determine, among the plurality of candidate text clusters, a candidate text cluster whose included text quantity satisfies a first quantity condition as a target text cluster, and events associated with multiple texts in the target text cluster belong to a target type event, and the target type event refers to an event whose associated text quantity satisfies the first quantity condition;

[0194] A weight determination module 1003, configured to determine weights of multiple texts in the target text cluster based on interaction information of the multiple texts in the target text cluster, and the weight of a text represents the importance degree of the text;

[0195] A text generation module 1004, configured to generate an event description text based on multiple texts in the target text cluster and weights of the multiple texts in the target text cluster, the semantics of the event description text is the same as the semantics of the multiple texts in the target text cluster, and the event description text is used to describe the target type event associated with the multiple texts in the target text cluster.

[0196] When the event determination device provided by the embodiment of the present application determines a target type event based on multiple texts associated with the event, it first clusters the multiple texts to obtain a plurality of candidate text clusters, and then screens out a target text cluster whose text quantity satisfies the first quantity condition from the plurality of candidate text clusters. Then, events associated with multiple texts in the target text cluster belong to the target type event. Then, using multiple texts in the target text cluster and weights of the multiple texts, an event description text of the event is generated. Since the importance degree of each text is considered, it can be ensured that the generated event description text focuses on important aspects. Therefore, although the processing method of the solution of the present application is simple and saves computing resources, it can still ensure the accuracy of the determined target type event. And by generating an event description text corresponding to the target text cluster, the user can clearly understand the target type event and multiple texts reflecting the target type event, improving the interpretability of the event determination method.

[0197] Optionally, the clustering module 1001 is configured to:

[0198] Extract features from multiple texts to obtain first semantic features of each text;

[0199] When the dimension of the first semantic feature of multiple texts is not greater than a preset dimension, cluster the multiple texts based on the first semantic feature of the multiple texts to obtain multiple candidate text clusters;

[0200] When the dimension of the first semantic feature of multiple texts is greater than the preset dimension, reduce the dimension of the first semantic feature of the multiple texts to obtain the second semantic feature of the multiple texts, and cluster the multiple texts based on the second semantic feature of the multiple texts to obtain multiple candidate text clusters.

[0201] Optionally, the clustering module 1001 is used to:

[0202] Extract keywords from multiple texts respectively to obtain at least one keyword in each text;

[0203] Determine at least one first target keyword among the keywords in multiple texts, and the number of occurrences of the first target keyword in multiple texts satisfies the second quantity condition;

[0204] For any first target keyword, determine all texts including the first target keyword in multiple texts as a candidate text cluster.

[0205] Optionally, the clustering module 1001 is used to:

[0206] Based on a preset multiple topic tags, identify multiple texts respectively to obtain the topic tag to which each text belongs;

[0207] Determine at least one target topic tag among multiple topic tags, and the number of texts under the target topic tag satisfies the third quantity condition;

[0208] For any target topic tag, determine all texts belonging to the target topic tag in multiple texts as a candidate text cluster.

[0209] Optionally, refer to Figure 11 , the apparatus further includes a text cluster merging module 1005, which is used to:

[0210] For each obtained candidate text cluster, determine the second target keyword of the candidate text cluster among the keywords in the texts included in the candidate text cluster, and the number of occurrences of the second target keyword in the texts included in the candidate text cluster satisfies the fourth quantity condition;

[0211] When the second target keywords of two candidate text clusters are the same, merge the two candidate text clusters into one candidate text cluster.

[0212] Optionally, refer to Figure 11 , the apparatus further includes a text cluster merging module 1005, which is used to:

[0213] Among the multiple candidate text clusters obtained, determine the text duplication degree between every two candidate text clusters, where the text duplication degree represents the proportion of the same text in the two candidate text clusters;

[0214] When the text duplication degree between two candidate text clusters is not less than the first preset threshold, merge the two candidate text clusters into one candidate text cluster.

[0215] Optionally, the weight determination module 1003 is used for:

[0216] For any text, obtain the interaction parameters in the interaction information of the text, where the interaction parameters include at least one of the number of forwards of the text, the number of likes of the text, the number of comments of the text, the number of views of the text, the number of favorites of the text, or the number of shares of the text;

[0217] Perform a weighted sum of the interaction parameters of the text to obtain the weight of the text.

[0218] Optionally, the text generation module 1004 is used for:

[0219] Determine multiple target texts among the multiple texts in the target text cluster, where the weights of the target texts meet the preset weight conditions;

[0220] Generate an event description text based on the multiple target texts, where the semantics of the event description text are the same as the semantics of the multiple target texts.

[0221] Optionally, the text generation module 1004 is used for:

[0222] Obtain a prompt text, where the prompt text is used to describe the goal that the expected generated event description text needs to meet;

[0223] Input the multiple target texts and the prompt text into a language processing model to obtain the event description text output by the language processing model.

[0224] Optionally, see Figure 11 , the multiple texts are obtained within the same period, and the target type events associated with the multiple texts in the target text cluster are target type events that occur within the period; the device further includes an event determination module 1006, which is used for:

[0225] For the first target text cluster and the second target text cluster obtained in different periods, determine the similarity between the first target text cluster and the second target text cluster;

[0226] When the similarity between the first target text cluster and the second target text cluster is not less than the second preset threshold, determine that the target type events associated with the multiple texts in the first target text cluster and the target type events associated with the multiple texts in the second target text cluster belong to the same target type event.

[0227] Optionally, refer to Figure 11 , the event determination module 1006 is configured to:

[0228] Aggregate keywords in multiple texts in the first target text cluster to obtain at least one first aggregated keyword;

[0229] Aggregate keywords in multiple texts in the second target text cluster to obtain at least one second aggregated keyword;

[0230] Determine the similarity between at least one first aggregated keyword and at least one second aggregated keyword as the similarity between the first target text cluster and the second target text cluster.

[0231] Optionally, refer to Figure 11 , the device further includes a display module 1007, which is configured to:

[0232] Display the event information of multiple determined target type events in the event display interface;

[0233] Wherein, the event information includes at least one of an event description text, texts in the target text cluster, the number of texts in the target text cluster, keywords in the texts in the target text cluster, the number of occurrences of the keywords, a new flag, or the number of occurrences of the target type event within a preset duration.

[0234] It should be noted that: for the event determination device provided in the above embodiment, only the above division of each functional module is used for illustration. In actual application, the above functions can be allocated to different functional modules according to needs, that is, the internal structure of the computer device is divided into different functional modules to complete all or part of the functions described above. In addition, the event determination device provided in the above embodiment and the event determination method embodiment belong to the same concept, and the specific implementation process is detailed in the method embodiment, which will not be repeated here.

[0235] The embodiment of the present application also provides a computer device, which includes a processor and a memory. At least one computer program is stored in the memory, and the at least one computer program is loaded and executed by the processor to implement the operations performed in the event determination method in the above embodiment.

[0236] Optionally, the computer device is provided as a terminal. Figure 12 The structural schematic diagram of a terminal 1200 provided by an exemplary embodiment of the present application is shown. The terminal 1200 includes a processor 1201 and a memory 1202.

[0237] The processor 1201 may include one or more processing cores, such as a quad-core processor, an octa-core processor, etc. The processor 1201 may be implemented in at least one hardware form of DSP (Digital Signal Processing), FPGA (Field Programmable Gate Array), or PLA (Programmable Logic Array). The processor 1201 may also include a main processor and a coprocessor. The main processor is a processor for processing data in the wake state, also known as the CPU (Central Processing Unit); the coprocessor is a low-power processor for processing data in the standby state. In some embodiments, the processor 1201 may be integrated with a GPU (Graphics Processing Unit), and the GPU is responsible for rendering and drawing the content to be displayed on the display screen. In some embodiments, the processor 1201 may further include an AI (Artificial Intelligence) processor, and the AI processor is used to process computational operations related to machine learning.

[0238] The memory 1202 may include one or more computer-readable storage media, and the computer-readable storage media may be non-transitory. The memory 1202 may further include high-speed random access memory and non-volatile memory, such as one or more disk storage devices and flash storage devices. In some embodiments, the non-transitory computer-readable storage media in the memory 1202 is used to store at least one computer program, and the at least one computer program is used to be possessed by the processor 1201 to implement the event determination method provided in the method embodiments of the present application.

[0239] In some embodiments, the terminal 1200 may further optionally include: a peripheral device interface 1203 and at least one peripheral device. The processor 1201, the memory 1202, and the peripheral device interface 1203 may be connected through a bus or signal lines. Each peripheral device may be connected to the peripheral device interface 1203 through a bus, signal lines, or a circuit board. Optionally, the peripheral device includes at least one of a radio frequency circuit 1204, a display screen 1205, a camera assembly 1206, an audio circuit 1207, and a power supply 1208.

[0240] The peripheral device interface 1203 can be used to connect at least one I / O (Input / Output) related peripheral device to the processor 1201 and the memory 1202. In some embodiments, the processor 1201, the memory 1202, and the peripheral device interface 1203 are integrated on the same chip or circuit board; in some other embodiments, any one or two of the processor 1201, the memory 1202, and the peripheral device interface 1203 can be implemented on a separate chip or circuit board, and this embodiment does not limit this.

[0241] The radio frequency circuit 1204 is used to receive and transmit RF (Radio Frequency) signals, also known as electromagnetic signals. The radio frequency circuit 1204 communicates with a communication network and other communication devices through electromagnetic signals. The radio frequency circuit 1204 converts an electrical signal into an electromagnetic signal for transmission, or converts the received electromagnetic signal into an electrical signal. Optionally, the radio frequency circuit 1204 includes: an antenna system, an RF transceiver, one or more amplifiers, a tuner, an oscillator, a digital signal processor, a codec chipset, a user identity module card, and so on. The radio frequency circuit 1204 can communicate with other devices through at least one wireless communication protocol. The wireless communication protocol includes but is not limited to: metropolitan area network, various generations of mobile communication networks (2G, 3G, 4G, and 5G), wireless local area network, and / or WiFi (Wireless Fidelity) network. In some embodiments, the radio frequency circuit 1204 may further include a circuit related to NFC (Near Field Communication), and this application does not limit this.

[0242] The display screen 1205 is used to display the UI (User Interface). The UI may include graphics, text, icons, videos, and any combination thereof. When the display screen 1205 is a touch display screen, the display screen 1205 also has the ability to collect touch signals on or above the surface of the display screen 1205. The touch signals can be input as control signals to the processor 1201 for processing. At this time, the display screen 1205 can also be used to provide virtual buttons and / or virtual keyboards, also known as soft buttons and / or soft keyboards. In some embodiments, there may be one display screen 1205, which is provided on the front panel of the terminal 1200; in other embodiments, there may be at least two display screens 1205, which are respectively provided on different surfaces of the terminal 1200 or are in a foldable design; in other embodiments, the display screen 1205 can be a flexible display screen, which is provided on the curved surface or the folding surface of the terminal 1200. Even further, the display screen 1205 can also be set to an irregular non-rectangular shape, that is, a special-shaped screen. The display screen 1205 can be prepared using materials such as LCD (Liquid Crystal Display) and OLED (Organic Light-Emitting Diode).

[0243] The camera assembly 1206 is used to collect images or videos. Optionally, the camera assembly 1206 includes a front camera and a rear camera. The front camera is provided on the front panel of the terminal 1200, and the rear camera is provided on the back of the terminal 1200. In some embodiments, there are at least two rear cameras, which are any one of a main camera, a depth-of-field camera, a wide-angle camera, and a telephoto camera respectively, to achieve functions such as the combination of the main camera and the depth-of-field camera to achieve the background blurring function, the combination of the main camera and the wide-angle camera to achieve panoramic shooting and VR (Virtual Reality) shooting functions, or other combined shooting functions. In some embodiments, the camera assembly 1206 may also include a flash. The flash can be a single-color temperature flash or a dual-color temperature flash. A dual-color temperature flash refers to the combination of a warm light flash and a cold light flash, which can be used for light compensation under different color temperatures.

[0244] The audio circuit 1207 may include a microphone and a speaker. The microphone is used to collect sound waves of the user and the environment, and convert the sound waves into electrical signals for input to the processor 1201 for processing, or input to the radio frequency circuit 1204 to enable voice communication. For the purpose of stereo collection or noise reduction, there may be multiple microphones, which are respectively arranged at different parts of the terminal 1200. The microphone may also be an array microphone or an omnidirectional collection microphone. The speaker is used to convert the electrical signals from the processor 1201 or the radio frequency circuit 1204 into sound waves. The speaker may be a traditional thin film speaker or a piezoelectric ceramic speaker. When the speaker is a piezoelectric ceramic speaker, it can not only convert electrical signals into sound waves audible to humans, but also convert electrical signals into sound waves inaudible to humans for uses such as ranging. In some embodiments, the audio circuit 1207 may further include a headphone jack.

[0245] The power supply 1208 is used to supply power to each component in the terminal 1200. The power supply 1208 may be alternating current, direct current, a disposable battery or a rechargeable battery. When the power supply 1208 includes a rechargeable battery, the rechargeable battery may support wired charging or wireless charging. The rechargeable battery may also be used to support fast charging technology.

[0246] Those skilled in the art can understand that Figure 12 the structure shown in

[0247] does not limit the terminal 1200, and may include more or fewer components than shown in the figure, or combine some components, or adopt different component arrangements. Figure 13

[0248] Embodiments of the present application further provide a computer-readable storage medium, in which at least one computer program is stored, and the at least one computer program is loaded and executed by a processor to implement the operations performed by the event determination method in the above embodiments.

[0249] The embodiment of the present application also provides a computer program product, including a computer program, which is loaded and executed by a processor to implement the operations performed by the event determination method in the above embodiment.

[0250] Those of ordinary skill in the art can understand that all or part of the steps in the above embodiment can be completed by hardware, or can be completed by instructing relevant hardware through a program. The program can be stored in a computer-readable storage medium. The storage medium mentioned above can be a read-only memory, a magnetic disk, an optical disk, etc.

[0251] The above are only optional embodiments of the embodiments of the present application, and are not intended to limit the embodiments of the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the embodiments of the present application shall be included in the protection scope of the present application.

Claims

1. An event determination method, characterized in that, The method includes: Clustering a plurality of obtained texts to obtain a plurality of candidate text clusters, each candidate text cluster including at least one text, where the text is a text associated with an event; Among the plurality of candidate text clusters, determining a candidate text cluster whose number of included texts meets a first quantity condition as a target text cluster, where the events associated with the multiple texts in the target text cluster belong to a target type event, and the target type event refers to an event whose associated text quantity meets the first quantity condition; Based on the interaction information of the multiple texts in the target text cluster, determining the weights of the multiple texts in the target text cluster, where the weight of the text represents the importance degree of the text; Based on the multiple texts in the target text cluster and the weights of the multiple texts in the target text cluster, generating an event description text, where the semantics of the event description text is the same as the semantics of the multiple texts in the target text cluster, and the event description text is used to describe the target type event associated with the multiple texts in the target text cluster.

2. The method according to claim 1, wherein The clustering the obtained multiple texts to obtain a plurality of candidate text clusters includes: Performing feature extraction on the multiple texts to obtain a first semantic feature of each text; When the dimension of the first semantic features of the multiple texts is not greater than a preset dimension, clustering the multiple texts based on the first semantic features of the multiple texts to obtain the plurality of candidate text clusters; When the dimension of the first semantic features of the multiple texts is greater than the preset dimension, performing dimensionality reduction on the first semantic features of the multiple texts to obtain second semantic features of the multiple texts, and clustering the multiple texts based on the second semantic features of the multiple texts to obtain the plurality of candidate text clusters.

3. The method according to claim 1, characterized in that The clustering the obtained multiple texts to obtain a plurality of candidate text clusters includes: Respectively performing keyword extraction on the multiple texts to obtain at least one keyword in each text; Determining at least one first target keyword among the keywords in the multiple texts, where the number of occurrences of the first target keyword in the multiple texts meets a second quantity condition; For any first target keyword, determining all the texts including the first target keyword in the multiple texts as a candidate text cluster.

4. The method according to claim 1, characterized in that The clustering the obtained multiple texts to obtain a plurality of candidate text clusters includes: Based on a plurality of preset topic tags, respectively identifying the multiple texts to obtain the topic tag to which each text belongs; Determining at least one target topic tag among the multiple topic tags, where the number of texts under the target topic tag meets a third quantity condition; For any target topic tag, determining all the texts belonging to the target topic tag in the multiple texts as a candidate text cluster.

5. The method according to claim 1, wherein The method further includes: For each obtained candidate text cluster, determining a second target keyword of the candidate text cluster among the keywords in the texts included in the candidate text cluster, where the number of occurrences of the second target keyword in the texts included in the candidate text cluster meets a fourth quantity condition; In the case where the second target keywords of two candidate text clusters are the same, merge the two candidate text clusters into one candidate text cluster.

6. The method according to claim 1, wherein The method further includes: Among the obtained multiple candidate text clusters, determine the text duplication degree between every two candidate text clusters, where the text duplication degree represents the proportion of the same text in the two candidate text clusters; In the case where the text duplication degree between two candidate text clusters is not less than a first preset threshold, merge the two candidate text clusters into one candidate text cluster.

7. The method according to any one of claims 1 to 6, characterized in that The determining the weights of multiple texts in the target text cluster based on the interaction information of the multiple texts in the target text cluster includes: For any text, obtain the interaction parameters in the interaction information of the text, where the interaction parameters include at least one of the number of forwards of the text, the number of likes of the text, the number of comments of the text, the number of views of the text, the number of collections of the text, or the number of shares of the text; Perform weighted summation on the interaction parameters of the text to obtain the weight of the text.

8. The method according to claim 1, characterized in that, The generating an event description text based on the multiple texts in the target text cluster and the weights of the multiple texts in the target text cluster includes: Determine multiple target texts among the multiple texts in the target text cluster, where the weights of the target texts meet a preset weight condition; Generate the event description text based on the multiple target texts, where the semantics of the event description text is the same as the semantics of the multiple target texts.

9. The method according to claim 8, wherein The generating the event description text based on the multiple target texts includes: Obtain a prompt text, where the prompt text is used to describe the goal that the expected generated event description text needs to meet; Input the multiple target texts and the prompt text into a language processing model to obtain the event description text output by the language processing model.

10. The method according to claim 1, wherein The multiple texts are obtained within the same period, and the target type events associated with the multiple texts in the target text cluster are target type events that occur within the period; The method further includes: For a first target text cluster and a second target text cluster obtained in different periods, determine the similarity between the first target text cluster and the second target text cluster; In the case where the similarity between the first target text cluster and the second target text cluster is not less than a second preset threshold, determine that the target type events associated with the multiple texts in the first target text cluster and the target type events associated with the multiple texts in the second target text cluster belong to the same target type event.

11. The method according to claim 10, wherein The determining the similarity between the first target text cluster and the second target text cluster includes: Aggregate the keywords in the multiple texts in the first target text cluster to obtain at least one first aggregated keyword; Aggregate the keywords in the multiple texts in the second target text cluster to obtain at least one second aggregated keyword; Determine the similarity between the at least one first aggregated keyword and the at least one second aggregated keyword as the similarity between the first target text cluster and the second target text cluster.

12. The method according to claim 1, characterized in that, The method further includes: In the event display interface, event information of multiple determined target type events is displayed; Among them, the event information includes at least one of the event description text, the text in the target text cluster, the number of texts in the target text cluster, the keywords in the texts in the target text cluster, the number of occurrences of the keywords, the new flag, or the number of occurrences of the target type event within a preset duration.

13. An event determination device, characterized in that, The device includes: A clustering module, configured to cluster multiple obtained texts to obtain multiple candidate text clusters, each candidate text cluster includes at least one text, and the text is a text associated with an event; A text cluster determination module, configured to determine, among the multiple candidate text clusters, the candidate text clusters whose included text quantity meets the first quantity condition as target text clusters, the events associated with the multiple texts in the target text clusters belong to target type events, and the target type event refers to an event whose associated text quantity meets the first quantity condition; A weight determination module, configured to determine the weights of the multiple texts in the target text cluster based on the interaction information of the multiple texts in the target text cluster, and the weight of the text represents the importance degree of the text; A text generation module, configured to generate an event description text based on the multiple texts in the target text cluster and the weights of the multiple texts in the target text cluster, the semantics of the event description text is the same as the semantics of the multiple texts in the target text cluster, and the event description text is used to describe the target type event associated with the multiple texts in the target text cluster.

14. A computer device, characterized in that, The computer device includes a processor and a memory, and at least one computer program is stored in the memory, and the at least one computer program is loaded and executed by the processor to implement the operations performed by the event determination method according to any one of claims 1 to 12.

15. A computer-readable storage medium, characterized in that, At least one computer program is stored in the computer-readable storage medium, and the at least one computer program is loaded and executed by a processor to implement the operations performed by the event determination method according to any one of claims 1 to 12.

16. A computer program product, comprising a computer program, characterized in that, The computer program is loaded and executed by a processor to implement the operations performed by the event determination method according to any one of claims 1 to 12.

Images